Real-time Dispatching and Relocation of Service Engineers
- Problem:
Given a service region consisting of several machines, repair base stations and service engineers.
Costs occur if a machine breaks down and a repairman does not reach it within the established time window.
The goal is to minimize costs by finding a smart policy to manage the service engineers. - Approaches used:
Compliance tables
Heuristic approach
Approximate Dynamic Programming
Markov Decision Theory - Results:
– Accurate estimation of the repair gives significant advantage only in case of large repair times.
– Compliance tables perform well only for the systems with small distances.
– The heuristic policy increases the fraction of calls answered in time in up to 60% compared to a static policy and compliance tables.
– The ADP approach outperforms all other approaches.
– If the repair times are large, the exact optimal policy is obtained from Markov Decision theory. - Practical advice:
1. In case of limited computational resources, use heuristic approach.
2. If there is one given system and the time is not very limited, use ADP approach to obtain high-performing policy.
3. In case of large repair times use the policy described in Section 9.2.
